Factors influencing water consumption in buildings in southern Brazil

This study aims to identify the factors influencing water consumption in residential buildings in the city of Joinville, in southern Brazil, through an analysis of socioeconomic variables of users and constructive variables of the building. Data were collected on water consumption, socioeconomic characteristics of the users and constructive characteristics of the buildings, from a sample dataset of 89 buildings including 3171 housing units. Two models describing consumption were identified from a multiple linear regression analysis by means of the stepwise method. The analysis showed there were several variables related to constructive aspects, including the alternative water supply system that were statistically significant in explaining both water consumption in buildings (in m(3)/month) and per capita water consumption (in liters per person per day). The model explains 87.41% of the water consumption in buildings in Joinville and 79.68% of the per capita consumption. After obtaining the models, water consumption was estimated on eight buildings that were not included in the sample dataset using two equations for water consumption in buildings and per capita water consumption. Leave-one-out cross validation complements the analysis. The two models feature reasonable explanatory and predictive capacity, displaying adjusted and predicted R-2 higher than 70%. The obtained results contribute to enhanced understanding of variables related to water consumption in residential buildings in southern Brazil, and may be useful to the authorities in developing sustainable policies. (C) 2018 Elsevier Ltd. All rights reserved.